ON A METHOD TO CALCULATE RISK FUNCTION IN A PROBLEM OF IDENTIFYING GROUPS OF MULTIVARIATE STUDENT DISTRIBUTIONS

摘要

Construction of decision rules for group and pointwise classification is an important problem of image recognition theory. Depending on the initial information optimal and other statistical rules may be constructed, but in both cases classification errors are unavoidable. In the present paper the problem of calculation of the risk function of group classification is solved in the case of multivariate Student t-distribution.